Skip to main content
Uses LanceDB’s built-in hybrid search, which combines the strengths of semantic and full-text search. Working against the BEIR dataset, it shows how searching for both the meaning of a query and the specific keywords in it produces more relevant results than either alone, and how a reranker refines what comes back.
This example is a runnable notebook in the vectordb-recipes repository rather than a page here.

Open in Colab

Run it end to end in the browser.

View on GitHub

Read the source.